Preliminary

Centralbankbalancesheetsandtheliquiditytrap

MichaelT.Kiley *

March2003 Version2.1

*Address : WhileonLeave(September2001-August2003) —OECDDepartmentCS1,2rue AndrePascal,75016ParisFRANCE. Email : Michael.Kiley@.org . PermanentAddress :MailStop67, FederalReserveBoard,Washington,DC20551. Email : [email protected] . Acknowledgements :Thisisasubstantiallyrevisedversionofanearlierpaper,“Escapingaliquiditytrap: Anold-fashionedanalysiswithanewtwist”.IwouldliketocolleaguesattheFederalReserveBoardand studentsatJohnsHopkinsUniversityinWashington,DCfordiscussionsthatprovidedthebaseforthis researchandDavidRomerforskepticalcommentsonanearlierdraft.Theviewsexpressedhereinare thoseoftheauthor,anddonotrepresentthoseoftheOECD,theFederalReserveBoard,ortheirrespective staffs.

i Centralbankbalancesheetsandtheliquiditytrap Abstract Theeffectivenessofpolicyinaliquiditytrapisconsideredinasimplemodel.Themodel illustratestheinteractionbetweenfiscalandmonetarypolicy.Ithighlightshow traditionalpolicyresponses,suchasexpansionaryopen-operationsor- financedtaxcuts,havenostimulative,butnon-traditionalfiscal/monetaryactions, suchas-financedtaxcutsornon-Ricardiantaxcuts,arestimulative.Thefiscal theoryoftheleveliscrucialinaliquiditytrap,asmoneylosesitsspecialcharacter oncethenominalratehitszero.Finally,theanalysishighlightstworesultsof practicalrelevanceregardingcentralbanklossesandtheexchangerateeffectsofpolicies inaliquiditytrap.Withregardtotheformer,themodelillustratesthatanymonetary actionthatliftstheeconomyfromtheliquiditytrapmustinvolveadeteriorationofthe centralbank’sbalancesheet;hence,aninstitutionalframeworkthatforbidssuchlosses hindersanescapefromaliquiditytrap.Withregardtothelatter,monetaryactionsthat succeedinliftingtheeconomyfromaliquiditytrapgeneratearealappreciation, reflectingthefiscalnatureofthemonetaryaction;hence,suchactionscannotbe criticizedon“beggar-thy-neighbor”grounds.

JELCodes:E52,E31,B22 Keywords:LiquidityTrap,,FiscalandMonetaryPolicyInteraction,Fiscal TheoryofthePriceLevel

ii 1. Introduction Previouslyrelegatedtothedustbinofhistory,theliquiditytrapisback–inmodernJapan andontheresearchagenda. 1Theappropriatepolicyresponsehasprovencontroversial. Suggestionsincludeantarget(Krugman(1998)),fiscalexpansion(Posen (1998)),aprice-leveltarget(Wolman(1998),Lilico(2002)),monetaryexpansion (Meltzer(1999)),ataxoncurrencyholdings(BuiterandPanigirtzoglou(1999)),currency depreciation(McCallum(1999),CoenenandWieland(2003)),andacombinationofa currencydepreciationandprice-leveltarget(Svensson(2001)).Mostoftheseproposals areatleastimplicitlydiscussedbelow,astheeffectsofdifferentmonetaryandfiscal actionsareexamined.Unfortunatelytheseproposalsmayhavelimitedeffectiveness,at leasttotheextentthatabaselinegeneralequilibriummodelwithmonetaryfrictions providesareasonableguideforpolicy. Themodeldevelopedhereinisatwo-period,deterministicopen-economymodel withanoptimizingrepresentativeconsumerandflexible.Inthemodel, expectations(or,morepreciselygivenperfectforesight,futurevalues)areveryimportant (forconsumptiondecisions,assetprices,andhencefortheliquiditytrap).Theresults demonstratethatthemonetaryauthority,throughopen-marketoperations,cannot engineerinflation,price-leveltargetsorcurrencydepreciationoncetheeconomyenters theliquiditytrap.ThissuggeststhatSvensson’s(2001)foolproofpolicycannotliftthe economyfromaliquiditytrapwithoutfiscalorbalancesheeteffectsashasbeenclaimed (apointdevelopedmorebelow). 2Thisresultisstandardforthetypeofmodelanalyzed, buthighlightsthatsuggestionsforsuchpoliciesmusteitherinvolvemonetaryactions otherthansecuritypurchasesorimperfectionsnotincludedinthemodel,suchas imperfectsubstitutabilitybetweendifferentassetsthatgeneratesportfoliobalanceeffects. Suchimperfectionsmaynotallowforverypowerfuleffectsofmonetarypolicyina liquiditytrap. 3

1Krugman(1998)getscreditforthesnappyphrasingandanearlyreturntothetopic. 2SomereadersmayimmediatelyobjectbeforeturningtotheanalysisthatSvensson(2001)doesnotrelyon portfoliobalanceeffects.However,acarefulreadingofhisappendix2suggestsotherwise. 3Lebow(1993)presentsacogentdiscussionofthelimitedrole“non-traditional”monetarychannels,such asportfoliobalanceorcreditchannels,mayhaveinaliquiditytrap,evenifsuchchannelsarepowerful outsidealiquiditytrap.

1 Traditionalfiscalpolicymayalsoberelativelyimpotentforseveralreasons. First,whenfiscalpolicyispursuedinaRicardianfashion(asdefinedbyWoodford (2001)orCochrane(1998)),theaggregateeffectsofdeficit-financedtaxcutsis nil(orverylimited)becauseofRicardianequivalence(Barro(1974)).Inaddition, governmentconsumptionandinvestmentmaybe(good)substitutesforprivatespending, implyingthatincreasesingovernmentspendingmayhavelittleeffecton(i.e.,asmallgovernmentspending(Krugman(1999))).Boththese reasonsmayexplainpartoftheapparentlyweakaggregatedemandeffectsoffiscal stimulusinJapaninthelate1990s(OECD(2000)).Anotherlimitoffiscalpolicymaybe theabilityofthefiscalauthoritytopursueexpansionarymeasuresgiventhelong-run fiscaloutlookandthefeasibilityofraisingfuturerevenuesorcuttingfutureexpenditures. Thisconcernmayhavebeenmoreimportantinlimitingthedegreeoffiscalexpansion pursuedbyJapanrecently(astheratioofgovernmentdebttoGDPexceeded120percent in2000andimplicitpublicpensionliabilitiesarelarge). Giventheimpotenceofsecuritiespurchasesorbond-financedtaxcutswhen facingaliquiditytrap,innovativemonetaryandfiscalpolicyactionsareexaminedinthe model.Monetaryactionstorelaxthegovernment’sbudgetconstraintornon-Ricardian fiscalpolicycanlifttheeconomyfromtheliquiditytrap.Theappealtothefiscaltheory ofthepriceleveliscrucial,asmoneylosesitsspecialcharacter(i.e.,anecessarybut dominatedassetclass)oncethenominalinterestratehitszero.Themonetaryandfiscal authoritiesmustcommittoinflatingaway(loosely,monetizing)someportionof outstandingnominalliabilities.Inessence,theeffectsofmonetaryactionsrepresent somepleasantmonetaristarithmetic–theboostingofinflationthroughthefiscaleffects ofmonetarypolicyemphasizedbySargentandWallace(1981).Itshouldbenotedthat relaxationofthegovernment’sbudgetconstraintthroughmonetaryexpansionisoften mentionedasapossibility(e.g.,Clouseetal(2000),andBernanke(2000)),buttypically isfrowneduponasinappropriate.Suchreasoningisdifficulttocomprehend,asmoney andbondsareperfectsubstitutesinaliquiditytrap;sinceconsumersarewillingto increasetheirholdingsofmoneyinaliquiditytrapandthegovernmentcanlowertax

2 burdensbyincreasingthesupplyofmoneywithoutresourcecosts,itwouldseem appropriateforthegovernmenttoissuemoremoneyandcuttaxes. 4 Twopracticalissuesarisefromtheanalysis.First,themonetaryactionssuggested havenegativeeffectsonthecentralbank’sbalancesheet,reflectingthemonetizationof governmentliabilities.Providingmoneytothetreasurywithoutanaccompanying increaseinassetsonthecentralbank’sbalancesheetimpliesthattheoutstandingstockof moneyexceedstheassetsheldbythecentralbank.Ofcourse,suchabalancesheet deteriorationisnoteconomicallyimportant;acentralbankcouldalwayssimplyprint moremoneywheneveranyonewishedtoexchangetheirmoneyforanasset(more money),withoutanychangeinthestockofmoneyheldbythepublic.However,such lossesarepreventedincertaininstitutionalsetups,mostnotablyinJapan,wheretheBank ofJapanisaprivateentitythatmustremainsolvent.Sinceonlymonetaryactionswith negativeconsequencesforthecentralbank’sbalancesheetallowanescapefromthe liquiditytrap,suchinstitutionalframeworkslimitthepotentialofmonetarypolicy. Inaddition,thetypeofmonetaryexpansiondescribedmayleadtoarealcurrency appreciationintheshortrun(ratherthantherealdepreciationpredictedbyportfolio balancechannels),reflectingthefiscalnatureofthemonetaryaction.Hence,suchpolicy actionscannotbecriticizedonthe“beggar-thy-neighbor”groundsofcallsforcurrency depreciation(e.g.,Stevens(2001),thediscussioninSvensson(2001)andCoenenand Wieland(2003)).Thistheoreticalpredictionhasnotbeenpreviouslydiscussed,and suggestsevenmorestronglythepotentialdesirabilityofthepolicyforJapanintheyears following2001,whenworldaggregatedemandappearstobefalteringandbeggar-thy- neighboractionsmayaggravateadversedevelopmentselsewhereinAsiaorintheUnited States( The (2001)).

2.Asimpleeconomy Themodelisofasmall,openeconomy,consistingofarepresentativeconsumeranda government(whichcanbebrokendownintoacentralbankandtreasuryifsucha

4Outsideoftheliquiditytrap,increasesinthemoneysupplyboostthenominalinterestrate(byincreasing inflation)andhencecreatedistortions.However,suchdistortionsareabsentintheliquiditytrap,wherethe nominalinterestrateiszero.

3 breakdownisdesired).Theeconomyconsistsoftwoperiods,period1andperiod2. Domesticoutputisexogenous,andthepricelevelisperfectlyflexible.Hence,the liquiditytrapisasituationwherethemonetaryauthoritycannotinfluencethepricelevel throughopen-marketoperations,ratherthanasituationwherethemonetaryauthority cannotinfluenceoutput.AsinKrugman(1998),addingstickypricesanddemand- determinedoutputlevelswouldtransformtheliquiditytraptoasituationwherethe centralbank’sattemptstoinfluenceoutputarethwarted,therebyinjectingaKeynesian flavorintotheanalysis.However,suchamodificationaddsnoinsightandhenceis ignoredforsimplicity. 2.1Therepresentativeconsumer Therepresentativeconsumerspreferencesoverconsumptionofthedomestically producedgood(c(j),j=1or2)andtheforeigngood(g(j),j=1or2)aregivenby

Eq.1 a ln[ c )]1( + 1( − a)ln[ g )]1( + D{a ln( c )2( + 1( − a)ln[ g )]2( ,} where“a”isaparameter(equaltotheexpenditureshareofdomesticintotal consumptionexpenditure)andDisthediscountrate.(Notethatinwhatfollows, variablesdenotedwithalower-caseletterrefertorealvariables,suchasconsumption, whilevariablesdenotedwithupper-caselettersarenominalvalues,suchasthe). Theconsumer’schoicesareconstrainedbyabudgetconstraintandatransaction technology.Theconsumerisendowedwithanincomeineachperiod(y),aninitialstock ofdomesticassets(bonds,B(1))andaninitialstockofforeignassets(F(1)).Thebudget constraintsinperiod1and2are

Q f F )2( g )1( F )1( Eq.2 M + QB )2( + + P )1( c )1( + = B )1( + + P )1( y − P )1( ty , E )1( E )1( E )1(

g )2( F )2( Eq.3 P )2( c )2( + = M + B )2( + + P )2( y − P )2( ty , E )2( E )2( whereM,B(2),andF(2)aretheconsumer’schoicesofmoney,domesticbond,and foreignbondholdingstobecarriedintoperiod2,QandQ f arethepurchasepriceof domesticandforeignbondsintheirowncurrency(andtheirinversesarethegross nominalreturnonthesebonds),P(j)isthepriceofdomesticgoodsinperiodj,E(j)isthe

4 nominalexchangerateinperiodj(i.e.,thevalueofforeigncurrencyintermsofdomestic currency,soanincreaseinE(j)isanominalappreciation),andtistaxrateonthe household’sendowment(whichisidenticalineachperiodforsimplicity).For convenience,theforeignpricelevelhasbeennormalizedtooneandisassumedtobe time-invariant,implyingthatthedomesticpriceofforeignconsumptiongoodsisthe inverseofthenominalexchangerate. Transactionsinperiod2aresubjecttoa-in-advanceconstraint;nosuch constraintexistsinperiod1.Thetransactiontechnologyis

g )2( Eq.4 P )2( c )2( + ≤ M . E )2( 2.2Thegovernment Thegovernmentissuesbondsandmoneyandcollectstaxes;asaconvenient simplification,governmentspendingisignored.Governmentspendingcouldbeaddedto themodelwithoutchangingtheresultsdiscussed;however,suchanadditionwould provideanavenuetoanalyzetheeffectsofgovernmentspendingontheequilibrium,and inparticularonanescapefromtheliquiditytrap.Thedetailsofsucheffectswouldlikely dependonthesubstitutabilityofgovernmentconsumptionforprivateconsumptionand thefinancingofadditionalspending(i.e.,whetherdistortionarytaxeswereneededor not).Asapracticalmatter,callsforfurtherincreasesingovernmentconsumption,at leastinJapan,havenotbeenpopular,asproductiveopportunitiesforsuchspending appearlimited,andhencesuchspendingisnotanalyzed. Thenotationforthedomesticstocksofgovernment-issuedbondsandmoney,and fortaxcollections,hasalreadybeenintroduced.Thegovernment’schoicesregardingM andtareexogenous(andB(1),whichhasbeendeterminedinsomeearlierperiod,isalso exogenous).Moreover,thespecificationassumesthattaxcollectionsareexogenousin realterms,notnominalterms,asthisbestreflectsactualpractice–wheretaxcollections aretypicallyafractionofincome,andhenceriseandfallwiththepricelevel.Theonly remainingaspectofthegovernmentisitssequenceofbudgetconstraints,ormore appropriatelythevaluationequationsforgovernmentdebt,whicharegivenby

5 Eq.5 M + QB )2( = B )1( − P )1( ty

Eq.6 B )2( = P )2( ty − M . Theleft-handsideofequation5isthenominalvalueofgovernmentliabilitiesissuedin period1,whichequalsthedifferencebetweendebtpaymentsandtaxrevenuesthat period.Equation6equatesnominalliabilitiestotaxcollectionsinperiod2.Notethat thesetwoequationscanbecombinedtoequatetheinitialstockofgovernmentnominal liabilities(B(1))tothepresentdiscountedvalueoftaxandseignoragerevenue B )1( = P )1( ty + QP )2( ty + 1( − Q)M . Seignoragerevenueisnon-zerowhenthenominalinterestrateexceedszero(i.e.,when Q<1).Thesequenceofbudgetconstraintsimpliesthatthevalueofgovernmentliabilities equalstheresourcestakenfromtheeconomytopayfortheliabilities.Asdiscussedin Cochrane(1998)orWoodford(2001),theequalitybetweenthenominalvalueof governmentdebtandthepresenttheresourcesusedtorepaythatdebtisanequilibrium conditionthatensuresthataninvestoriswillingtoholdgovernmentdebt,notaconstraint onthefeasiblesetofgovernmentpolicies,atleastundercertainconditions. 2.3Theforeignsector Theeconomyisasmallopeneconomythattakestheworldinterestrate(1/Q f-1) asgiven.Therest-of-the-worldinteractswiththedomesticeconomysolelyinproviding importsoftheforeigngood(elasticallyatthenominalexchangerate)andforeignbonds asassets;itdoesnotdemandanyexportsfromthedomesticeconomy,againaconvenient simplification. Giventhis,theonlyconstraintsimposedonthedomesticeconomybytheforeign sectorarebalanceofpaymentsconstraints.Theseequatetheinitialstockofthe economy’sforeignassetstothepresentdiscountedvalueofimports(attheworldreal interestrate,whichequalsthenominalrateastheworldpricelevelisfixed)andaregiven by

6 Eq.7 F )1( = Q f F )2( + g )1(

Eq.8 F )2( = g )2( . Aspurchasesofforeigngoodscanonlyoccuriftheinitialstockofforeignassetsis positiveandpositivepurchasesarenecessaryfortobebounded,theexogenous initialstockofforeignassetsisrestrictedtobegreaterthanzero. 2.4Equilibrium Theequilibriumischaracterizedbythevaluesofconsumption(domesticand foreign),prices,anddomesticinterestrates(c(j),g(j),P(j),E(j),andQforj=1and2) thatmaximizethevalueoftherepresentativeconsumer’spreferences(eq.1)subjectto thebudgetconstraintsandtransactiontechnology(eqs.2to4)andthatareconsistent withtheresourceconstraint,governmentdebtvaluationequationsandbalanceof paymentsconstraints(eqs.5through8)giventheexogenousvaluesforoutput, governmentdebtandtaxes,themoneysupply,foreignassets,andtheworldreal(y,B(1),t,M,F(1),andQ f).Theequilibriumcanbecharacterizedanalytically,as showninanappendix.Suchananalyticalcharacterizationisparticularlyvaluablewhen examiningtheliquiditytrap,whichoccurswhenthenominalinterestrateequalszeroand hencethecash-in-advanceconstraint(eq.4)doesnotbind.Itisoftendifficultto analyticallycharacterizeequilibriumatsuchcornersolutionsinmorecomplicatedor many-periodmodels,andhencemoretransparentresultsarefoundinthesimplesetup herein. 2.4Thecentralbankandthetreasury Thediscussionsofarhasignoredanydistinctionbetweenthecentralbankandthe treasury,asonlytheconsolidatedgovernmentpositionaffectsequilibrium–given governmentpoliciesonthemoneysupply,taxesandtheinitialsupplyofgovernment debt.However,adistinctionbetweenthecentralbankandtreasuryisusefulwhen consideringthesetofpoliciesconsistentwithinstitutionalconstraintsonthecentralbank. Inparticular,itiscommonforalegalrequirementtoexistthatlimitstheoutstanding supplyofmoney(M)–theliabilityofthecentralbank–tobelessthanorequaltothe centralbank’sassetholdings,andforanysurplusincome(seignorage)tobereturnedto thetreasury.Assumingthatthecentralbank’sopenmarketoperationsinvolvethesale

7 andpurchaseofgovernmentdebt(B(cb))andthatthisconstraintisbinding,suchpolicies implythefollowingconditions

Eq.9 M = B(cb )

Eq.10 Seignorage = B(cb ) − QB (cb ) = 1( − Q)M . Theeffectoftheseconstraintsonthetreasuryhasalreadybeenincorporatedinthe governmentbudgetconstraints(equations5and6).Oneimportantimplicationofthese constraintsisthatthecentralbankcanonlyaffectthegovernment’sbudgetposition outsidealiquiditytrap(whenQ<1).Thisimpliesthatmoney-financedtaxcuts(or purchasesofgoods)areinadmissible,asthemoneysupplysocreatedwouldnotbe backedbygovernmentdebt.

3.Theliquiditytrap Thefactorsinfluencingthelikelihoodofaliquiditytrapandthenatureoftheeconomy’s responsetochangesingovernmentpoliciesaresimpletodescribe;thedescriptionsare providedasasetofproperties.Proofsofthepropertiesarerelegatedtoanappendix. First,adefinition:a liquiditytrap occurswhenthenominalinterestrateequals zero(orQequalsone).Theforcesdeterminingwhethersuchtrapwilloccurandthe behavioroftheeconomyaresummarizedbyfiveproperties.Thefirstpropertyisa specialfeatureofthemodelthatcontributestoitstractability: Property1: Inflationisentirelydeterminedbytime(P(2)/P(1)equals D). Asthisresultimpliesafixedconstantofproportionalitybetweenthefirstandsecond periodpricelevels,theremainderofthediscussionsimplyreferstothepricelevel. Thenexttwopropertiesfocusontheimpactofmonetaryandfiscalactionsonthe pricelevelandnominalinterestrates. Property2: Thenominalinterestrateisdecreasinginthemoneysupply(M), decreasinginthetaxrate(t)andincreasingintheinitialstockofdebt(B(1)) outsidetheliquiditytrap. Eachoftheseeffectsisstandardfromundergraduatetreatments–loosemonetarypolicy lowersnominalinterestrates,asdoesatightfiscalpolicy.However,inthepresentcase

8 theseeffectsarisethroughtheinteractionoffiscalandmonetarypolicyonthevaluation ofgovernmentdebt.Thegovernment’sbudgetconstraintimplies

Eq.11 B )1( = P )1( ty + QP )2( ty + 1( − Q)M . Lowerlevelsofdebt,highertaxratesandalargermoneysupplyallimplylessneedfor seignoragerevenue,andhenceincreaseQ(orlowerthenominalinterestrate)–untilthe nominalinterestratehitszeroandtheeconomyenterstheliquiditytrap.Figure1graphs thenominalinterestrateagainstthemoneysupply,illustratingthenegativerelationship. Thedashedlineinthefigureillustratestheimpactofataxcut–whichshiftsoutthelocus summarizingthenominalinterestrate/moneysupplyrelationship,increasingthenominal interestrateforagivenmoneysupply. Property3: Thepricelevelisincreasinginthemoneysupply(M)outsidethe liquiditytrap.Moreover,themoneysupplyistheonlyvariableunderthe government’scontrolthatinfluencesthepriceleveloutsidetheliquiditytrap,i.e. thepricelevelis moneydetermined . Outsidealiquiditytrap,thecash-in-advanceconstraint(equation4)isbindingandthe pricelevelfollowsstandardquantitytheorylogic,i.e.,thepricelevelisproportionalto themoneysupply. Theinterestingpropertiesofthemodelariseinaliquiditytrap. Property4: Oncetheeconomyentersaliquiditytrap,furtherincreasesinthe moneysupplyhavenoeffectonanyendogenousvariable.Thenominalinterest ratecanbeliftedabovezerobytaxcuts.Thepricelevelremainsstrictly decreasingintaxrateswithintheliquiditytrap,andissolelydeterminedbyfiscal variables(B(1)andt),i.e.,thepricelevelis fiscallydetermined . Theinabilityoffurtherincreasesinthemoneysupplytoaffectnominalinterestratesor thepricelevelarestandardresultsoncethezeroboundonnominalinterestratesare reached.Thisagainreflectstheinfluenceofthegovernment’sbudgetconstrainton equilibrium;oncethenominalinterestratehitszero(Qequals1),furtherincreasesinthe moneysupplyhavenoimpactofgovernmentbudgetbalance(equation9)orhousehold decisions(asmoneyandbondsareperfectsubstitutes).Theswitchtoa fiscally determined equilibriumismoreinterestingandappealstothefiscal-theoryoftheprice level(Cochrane(1998)andWoodford(2001)).Oncemoneybecomesirrelevant,theprice

9 levelissolelydeterminedbyrequirementsforfiscalbalance.Taxcutsincreasetheprice level(andtaxincreaseslowerthepricelevel),assuchmovementsarenecessarytoensure thatthenominalvalueofgovernmentliabilitiesequalsnominaltaxcollections.Of course,thetypeoftaxmovementsconsideredare non-Ricardian ;taxcutsoneperiodare notoffsetbytaxcutsintheotherperiod.Rather,ataxcutisfinancedbyanincreasein thepricelevel,boostingnominaltaxcollections. Figures1and2illustratetheeffectsoffiscalpolicychangesinaliquiditytrap. Consideraninitialequilibriumwheretheeconomyisinaliquiditytrap,butonlybarely– sothemoneysupplyisjusttotherightofM*.Atthispoint,anyfurtherincreasesinthe moneysupplyhavenoeffectonthepricelevelornominalinterestrates.Ataxcutshifts thelocussummarizingthenominalinterestrate/moneysupplyrelationshiptotheright, boostingthenominalinterestratesolongasthemoneysupplyislessthanM**. Moreover,asshowninfigure2,ataxcutbooststhepricelevelregardlessoftheposition ofthemoneysupplybyshiftingupthepricelevel/moneysupplylocusovertheliquidity trapregionwherethepricelevelis fiscallydetermined .Atlowerlevelsofthe(outsidetheliquiditytrap)–i.e.,totheleftofM*–fiscalpolicyhasnoimpacton thepricelevel,asthepricelevelis moneydetermined . Corollaryofproperty4: Conventionalexpansionaryopenmarketoperations havenoeffectontheequilibriumoncetheeconomyisinaliquiditytrap. Monetaryactionswithdirectfiscalconsequences,suchasmoney-financedtax cuts,canlifttheeconomyfromtheliquiditytrapandraisethepricelevel,but involvecreationofmoneyinexcessoftheassetsonthecentralbank’sbalance sheetandhencearelegallyprohibitedbyregimeswithconstraintslikeequation9. Thisresultstemdirectlyfromproperty3andtheconventionalconstraint(9)thatmoney isinjectedthroughopenmarketoperations(eitherpurchasinggovernmentbondsorother assetsthatinthecurrentmodelareperfectsubstitutesforgovernmentbondsinaliquidity trap).Inparticular,theincreaseinthemoneysupplynecessarytofinanceataxcutwill leavetheeconomybetweenM*andM**inthemodel(becausetheincreaseinthe moneysupplyneededislessthantheincreasetoM**),implyingapositivenominal interestrateandhigherpricelevel.Additionalincreasesinthemoneysupply,abovethe

10 levelrequiredforthetaxcut,wouldraisethepricelevelfurtheruntilthemoneysupply exceedsM**. Thefinalpropertypresenteddiscussestheimpactofmoneysupplyandmovementsonthenominalandrealexchangerate. Property5: 1.Outsidealiquiditytrap,increasesinthemoneysupplycausea nominalexchangeratedepreciation(intheinitialperiod,i.e.,lowerE(1)).2.The nominalexchangerateinperiodoneisdecreasinginthetaxrateandincreasingin theinitialstockofforeigndebt,i.e.,tighterfiscalpolicyleadstoarealand nominaldepreciationinperiod1.3.Uncoveredinterestparityholds,implyingthat thedegreeofnominalexchangeratedepreciationbetweenperiodsoneandtwois higherthehigheristhenominalinterestrate–andhencethetighteris(lowerM)andthelooserisfiscalpolicy(higherB(1)andlowert).4.Ina liquiditytrap,increasesinthemoneysupplyhavenoeffectonthenominalorreal exchangerate.5.Inaliquiditytrap,amoney-financedtaxcut(ornon-Ricardian taxcut)generatesarealexchangerateappreciation(orunchangedrealexchange rate)inperiodone. Forthemostpart,theseresultsreflectstandardforcesandoperatethroughuncovered interestparity.However,thelastresult–thatamoney-financedtaxcutgeneratesareal exchangerateappreciation–isnotoftenemphasized.Inparticular,thisresultstemsfrom thefiscalaspectofthemonetaryexpansion(thetaxcut),asthemonetaryaspectisabsent whenthecash-in-advanceconstraintisnotbindingandthepricelevelandnominal interestrateare fiscallydetermined . Figure3illustratesthesepoints.Outsidetheliquiditytrap(totheleftofM*),the realexchangerateP(1)E(1)isdecreasinginthemoneysupply.Thisoccursbecauseat higherlevelsofthemoneysupply,thenominalinterestrateislower,reducingthetaxon consumptionofforeigngoodsinthesecondperiod(g(2))thatarisesfromthecash-in- advanceconstraint.Thedecreaseinthistaxlowersdemandfortheforeigngoodinperiod one,exertingdownwardpressureontherealexchangerate.Thedashed-lineillustrates theeffectofafiscalexpansion(taxcut)financedthroughmoneycreation.Thisshiftsthe realexchangerate/moneysupplylocusupward,astheresultinghighernominalinterest rateincreasesthetaxonconsumptionoftheforeigngoodinperiod2throughthecash-in-

11 advanceconstraintandhenceincreasesperiod-onedemandfortheforeigngood,exerting upwardpressureontherealexchangerate.Thehighernominalinterestratefollowsfrom thecorollarytoproperty4,i.e.,thattheadditionalmoneynecessarytofinanceataxcut leavestheeconomybelowM**.

3.Discussionandcaveats Themodelemphasizestheimportanceoffiscaleffectsofmonetarypolicy.Inthisrespect theanalysisechoeslessformaldiscussions(i.e.,outsideanexplicit,optimizingmodel)of theliquiditytrap.Theresultsstemfromthefactthatseignoragerevenuesonthe outstandingstockofmoneyarezeroinaliquiditytrap,reflectingthezeronominal interestrate.Inthissituation,anytransfertothetreasury–oranyfiscaleffectsof monetaryactions–involveaworseningofthecentralbank’sbalancesheet.Some observershavenotedthat,inpractice,itmaythereforemakesensetoallowacentralbank tohavenegativenetworth.ThisissuggestedinBernanke(2000)andcurrentlyisnot allowedinmostcountrieswheremonetaryexpansionsmustoccurthroughpurchasesof assetsforthecentralbank’sbalancesheet,e.g.theUnitedStates(Clouseetal.(2000)) andJapan(OdaandOkina(2000)). OnerecentcontributionwithasimilarflavoristhatofEggertson(2001),who emphasizestheneedforancoordinatedfiscal/monetaryexpansion.Onenotable differenceistheinclusionofthecentralbank’sbalancesheetintheanalysis:Inorderto pursueanexpansionaryfiscal/monetarymixinaliquiditytrap,thecentralbankmust createunbackedmoney.Eggertsondoesnotconsiderwhetherthisconstraintoperatesin hismodel,ignoringanimportantinstitutionalfeature. TheanalysisalsoshedslightonthefoolproofescapemethodofSvensson(2001), whichconsistsofaninitialcurrencydepreciationaccompaniedbyaprice-leveltarget. Svenssondoesnotformallyconsiderwhethersuchapolicyisimplementablethrough open-marketoperations,butsuggestsinanappendixthatsuchaconstraintonbehaviorisunimportant.Byexplicitlymodelingtheinteractionoffiscaland monetarypolicyinanoptimizingmodel,thecurrentanalysisindicatesthatthe“foolproof way”maynotbesofoolproof,asitisnotimplementableinthecurrentmodel.Itisalso interestingtonotethatamoney-financedtaxcutleadstoarealexchangerate

12 appreciationinthemodel,notadepreciation.Thistheoreticalpredictionsuggestseven morestronglythepotentialdesirabilityofthepolicyforJapanintheyearsfollowing 2001,whenworldaggregatedemandappearstobefalteringandbeggar-thy-neighbor actionsmayaggravateadversedevelopmentselsewhereinAsiaorintheUnitedStates (seeStevens(2001),theothercommentsonSvensson(2001), TheEconomist (2001)or CoenenandWieland(2003)). Ofcourse,onestronglimitationofthecurrentanalysisisthetwo-period framework,whichseverelylimitsthedynamicaspectsofthemodel(asshowninproperty one,whereinflationisentirelydrivenbythediscountrate).However,theresultsare moregeneralthantheyappear.Consideraninfinite-horizonmodel,andsupposethat governmentdebtconsistssolelyofone-periodnominaldebt.Inperiodt,thevalueof suchdebt(paymentstothepublic)isB(t).Thestockofmoneyheldbythepublicatthe beginningofperiodt,issuedinthepreviousperiod,equalsM(t).Further,inperiodt,the amountofdebtcarriedintothenextperiodischosenbythetreasury(B(t+1)),andthe priceofsuchdebtinperiodtisQ(t)(andisdeterminedinthebondmarket).Denote nominaltaxcollectionsinperiodtbyT(t)andnominalgovernmentpurchasesbyG(t). Thevaluationequationforthegovernment’snominaldebt,reflectingtheconsolidationof treasuryandcentralbankaccounts,is

Eq.12 Q(t)B(t + )1 = B(t) + G(t) − T (t) − M (t + )1 + M (t) . Iteratingequation12forward(andimposingthenecessaryconditionthatthereal presented-discountedvalueofgovernmentdebtnotdivergeineitherdirection,i.e.,a transversalitycondition)yields

∞ j Eq.13 B(t) + M (t) = ΠQ(t + k){ T(t + j) − G(t + j) + 1( − Q(t + k + ))1 M (t + j + })1 å k=0 j=0 Nowconsiderdifferenttypesofmonetaryactionsinaliquiditytrapwherethenominal interestrateis(expectedtobe)equaltozeroovertheindefinitehorizon.Itisclearfrom equation13thatnosequenceofmonetaryactionshasanyimpactontheequilibrium,as thegovernment’sbudgetconstraintisunaffected(andbondsandmoneyareperfect substitutesinhouseholds’balancesheets). Theimportanceoffiscaleffectsofmonetarypolicycanbeextendedtoconsider theeffectsofalternativemonetaryactions.Forexample,considerthepurchaseof

13 privately-issuedsecuritiesviamoneycreation.Suchpurchasesareoftensuggestedas onerouteoutofaliquiditytrap(e.g.,Clouseetal(2000)).Ofcourse,inabaselinemodel withoneinterestrate,privatesecuritiesareperfectsubstitutesforgovernmentsecurities, andsuchopen-marketoperationshavenoeffects.Imperfectsubstitutability,ora portfoliobalancechannel,isonewayinwhichsuchpurchasescouldhaverealeffects. However,absentsucheffects,whichareoftenconsideredtobequitesmall(Lebow (1993)),fiscaleffectsofmonetarypolicyactionsarecriticaltoescapingtheliquiditytrap viamoneycreation. BothKrugman(1998)andAuerbachandObstfeld(2003)emphasizethe importanceofcreatingexpectationsoffutureexpansioninthemoneysupply.Their intuitionflowsfromequation13.Inparticular,ifshort-termnominalinterestratesare expectedtobepositiveatsomepointinthefuture,anincreaseinthemoneysupplyat thatpointrelaxesthegovernment’sbudgetconstraintandhencecaninfluencethecurrent equilibrium.Asidefromthedifficultiesinengenderingsuchexpectations,itisimportant tonotethatevensuchpoliciescouldresultinadverseconsequencesforacentralbank’s balancesheetandhencemaynotbepursuedinpractice.Forexample,supposethatlong- terminterestratesimplysomeexpectationthatshort-termnominalinterestrateswillbe positiveatsomedatet+kinthefuture.Inaddition,supposethatthecentralbank’sassets consistofsomemedium-termgovernmentbonds(horizonlessthank).Expectationsofa monetaryactioninthefuturecanaffecttoday’sequilibriumviaequation13.Ifsuch effectslifttheeconomyfromtheliquiditytrapinthemediumterm,i.e.,raiseexpected short-termratesatahorizonlessthank,thiswilllowerthevalueofgovernment securitiesonthecentralbank’sbalancesheet(asthevalueofthebondstodayis decreasingintheirexpectedyieldforagivenfacevalue).Suchadversemovementsinthe valueofcentralbankassetswouldrequireamonetarycontractiontomaintainapositive net-worthposition.Thismaymakeitverydifficult(inpractice)tocreateexpectationsof amonetaryexpansion.

5.Summary Aliquiditytrapisaproblemforconventionalstabilizationpolicies:Openmarket operationsareimpotent,andRicardianequivalenceandthelong-runsustainabilityof

14 fiscalpolicylimittheimpactoftraditionalfiscalpolicy.Becauseoftheseproblems, therehasbeenacallfortheuseofallemergencymeasures–purchaseofprivate securities,interventioninforeignexchangemarkets,implementationofinflationtargets– inthehopethatsomethingwillwork(Svensson(1999)). Theanalysishereinprovidesasimpledescriptionofpolicyoptionsandtheir theoreticaleffects.Themodelillustratestheinteractionbetweenfiscalandmonetary policy.Ithighlightshowtraditionalpolicyresponses,suchasexpansionaryopen-market operationsorbond-financedtaxcuts,havenostimulativevalue,butnon-traditional fiscal/monetaryactions,suchasmoney-financedtaxcutsornon-Ricardiantaxcuts,are stimulative.Thefiscaltheoryofthepriceleveliscrucialinaliquiditytrap,asmoney losesitsspecialcharacteroncethenominalinterestratehitszero.Finally,theanalysis highlightstworesultsofpracticalrelevanceregardingcentralbanklossesandthe exchangerateeffectsofpoliciesinaliquiditytrap.Withregardtotheformer,themodel illustratesthatanymonetaryactionthatliftstheeconomyfromtheliquiditytrapmust involveadeteriorationofthecentralbank’sbalancesheet;hence,aninstitutional frameworkthatforbidssuchlosseshindersanescapefromaliquiditytrap.Withregardto thelatter,monetaryactionsthatsucceedinliftingtheeconomyfromaliquiditytrap generatearealcurrencyappreciation,reflectingthefiscalnatureofthemonetaryaction; hence,suchactionscannotbecriticizedonthe“beggar-thy-neighbor”grounds. Inemphasizingfiscaleffects,portfoliobalancechannelsneednotbe quantitativelyimportant.Withthatsaid,theindependenceofmostcentralbanksin moderneconomiessuggeststhatsuchpoliciesmaybedifficulttoimplementwithout coordinationwithfiscalauthorities.Suchlegalandpracticalissueshavealreadybegun attractingresearch(seeClouseetal(2000)foradiscussionoflegalissuesintheUnited States,andEggertsson(2001)forananalysisofthetime-consistencyoffiscal/monetary coordination).

15 Appendix DenotingtheLagrangemultipliersonconstraints2,3and4facingthehousehold byw(1),w(2)andw(3),respectively,andtakingderivativesyieldsthefollowingfirst- orderconditionsforthehouseholdschoicevariables(c(j),g(j),M,B(2),F(2),j=1,2)

a A1 = P )1( w )1( c )1(

a A2 D = P )(2( w )2( + w ))3( c )2(

1− a w )1( A3 = g )1( E )1(

1− a w )2( + w )3( A4 D = g )2( E )2(

A5 w )1( = w )2( + w )3(

A6 Qw )1( = w )2(

Q f w )1( w )2( A7 = . E )1( E )2(

Combiningtheseexpressionswiththeresourceconstraints(c(1)=c(2)=y),thegovernment budgetconstraints(equation5and6),thebalanceofpaymentsconstraints(equation7 and8)andthecash-in-advanceconstraint(equation4,withappropriatecomplementary slackcondition,sothatthisconstraintbindswhenw(3)isgreaterthanzero)yieldsthe equilibriumexpressionsforpricelevels,thenominaldiscountfactor(andnominal interestrate)andexchangerates

a M B )1( A8 P )1( = min[ , ] D y t 1( + D)

M DB )1( A9 P )2( = min[ a , ] y t 1( + D)

B )1( a 1− + t A10 Q = max[ ,1 M D ] 1− at

16 a 1 F )1( A11 E )1( P )1( = 1− a 1+ DQ y

Q P )2( A12 E )2( P )2( = E )1( P )1( . Q f P )1(

Thepropertiesdiscussedinthetextfollowdirectlyfromtheseexpressions.NotethatA 12isanuncoveredinterestparitycondition,andthefirstpartsofA8andA9reflectthe quantityequationreasoningthatderivesfromthebindingcash-in-advanceconstraint.

17 References Auerbach,AlanandMauriceObstfeld(2003)TheCaseforOpen-MarketPurchasesina LiquidityTrap.Mimeo,UniversityofCaliforniaatBerkeley,February. Barro,RobertJ.(1974)AreGovernmentBondsNetWealth? Journalof 81:1095-1117. Bernanke,BenS.(2000)JapaneseMonetaryPolicy:ACaseofSelf-InducedParalysis?In Posen,AdamandRyoichiMikitani, Japan'sFinancialCrisisandItsParallelsto USExperience. InstituteforInternationalEconomics,Washington,DC,149-166. Buiter,WillemH.andNikolaosPanigirtzoglou(1999)LiquidityTraps:HowtoAvoid ThemandHowtoEscapeThem. NBERWorkingPaper7245 . Clouse,Jamesetal.(2000)MonetaryPolicyWhentheNominalShort-TermInterestRate IsZero.FederalReserveBoard, FederalReserveBoardFinanceandEconomics DiscussionPaper 2000-51. Cochrane,John(1998)AFrictionlessViewofInflation. NBERAnnual 323-384. Coenen,GunterandVolkerWieland(2003)TheZero-Interest-RateBoundandtheRole oftheExchangeRateforMonetaryPolicyinJapan.Mimeo,GoetheUniversityof Frankfurt,March. Eggertsson,Gauti(2001)CommittingtoBeingIrresponsible:DeficitSpendingtoEscape aLiquidityTrap.Mimeo,PrincetonUniversity. Krugman,PaulR.(1998)It’sBaaack!Japan’sSlumpandtheReturnoftheLiquidity Trap. BrookingsPapersonEconomicActivity 2. Krugman,PaulR.(1999)TimeontheCross;CanFiscalPolicySaveJapan?Mimeo. Lebow,DavidE.(1993)MonetaryPolicyatNear-ZeroInterestRates.FederalReserve Board EconomicActivityWorkingPaperSeriesNumber136. Lilico,Andrew(2002)TheLiquidityTrapandPrice-LevelTargeting. EconomicAffairs 22:47-52. McCallum,BennettT.(1999)TheoreticalAnalysisRegardingaZeroLowerBoundon NominalInterestRates.Mimeo,CarnegieMellonUniversity.

18 Meltzer,AllanH.(1999)Commentary:HowShouldMonetaryPolicyBeConductedin anEraofPricesStability?In NewChallengesinMonetaryPolicy ,BankofKansasCitySymposium. Oda,NobuyukiandKunioOkina(2000)FurtherMonetaryEasingPoliciesUnderthe Non-negativityConstraintsofNominalInterestRates:Summaryofthe DiscussionBasedonJapan’sExperience.InstituteforMonetaryandEconomic Studies,BankofJapan, IMESDiscussionPaperSeries 2000-E-24. OECD(OrganisationforEconomicCooperationandDevelopment)(2000) Economic SurveyofJapan1999/2000. Paris. Posen,AdamS.(1998) RestoringJapan’sEconomicGrowth .Institutefor,Washington,DC. Sargent,ThomasJ.andNeilWallace(1981)SomeUnpleasantMonetaristArithmetic. FederalReserveBankofMinneapolisQuarterlyReview 5(3). Stevens,Glenn(2001)CommentonSvensson(2001). MonetaryandEconomicStudies 19(S-1):313-316. Svensson,LarsE.O.(1999)Commentary:HowShouldMonetaryPolicyBeConductedin anEraofPricesStability?In NewChallengesinMonetaryPolicy ,Federal ReserveBankofKansasCitySymposium. Svensson,LarsE.O.(2001)TheZeroBoundinanOpen-Economy:AFoolproofWayof EscapingfromaLiquidityTrap. MonetaryandEconomicStudies 19(S-1):277- 312. TheEconomist (2001)Deflation.November17,82-83. Woodford,Michael(2001)FiscalRequirementsforPriceStability. NBERWorkingPaper 8072.

19 Figure1:Thenominalinterestratei

140.00

120.00

100.00

80.00 i 60.00

40.00 effectoftaxcut 20.00 initialequilibrium 0.00 0.00 1.00 2.00M* 3.00 M** 4.00 5.00 M

20 Figure2:ThepricelevelP

4.00 effectoftaxcut

3.50

3.00

2.50

P 2.00

1.50 initialequilibrium 1.00

0.50

0.00 0.00 1.00 2.00M* 3.00 M** 4.00 5.00 M

21 Figure3:TherealexchangerateE(1)P(1)

1.80 1.70 1.60 1.50 1.40 E(1)P(1) 1.30 1.20 effectoftaxcut 1.10 1.00 0.90 initialequilibrium 0.80 0.00 1.00 2.00M* 3.00 M** 4.00 5.00 M

22